Verizon's chief technology officer says the wireless industry is "missing a breakthrough" in 6G uplink capacity, as AI applications threaten to push traffic ratios from today's 9:1 downlink bias toward something closer to 1:1.

Yago Tenorio issued the warning at RCR Wireless' 6G Forum, where he argued that neither incremental engineering nor today's network architecture will cope with devices uploading sensory data to cloud-based AI agents in real time.

Mobile networks have spent two decades allocating most of their bandwidth to downloads, on the reasonable assumption that users consume far more than they transmit. Typical downlink-to-uplink ratios run at 9:1 or 8:2. Verizon expects AI to break that assumption.

What changes when AI goes continuous?

Verizon's scenario is a shift away from transactional AI prompts and answers toward what Tenorio called "continuous contextual awareness" — a device streaming information about a user's reality through sensing, video or audio, on a constant basis, to "your AI agent of choice" in the cloud or at the network edge.

The problem compounds on the device side. Future AI wearables, such as smart glasses, face hard limits on size, power, heat dissipation and the number of antennas that can realistically be crammed in.

"Everything puts pressure on the uplink. You need more capacity, but the device is less capable," Tenorio said.

Some techniques can help. Tenorio pointed to sub-band full duplex as one option. But his broader message was that the industry needs bigger thinking to avoid what he framed as a looming "uplink crisis."

"We're missing a breakthrough that will probably have to go all the way down to the architecture…There is a very real uplink demand that I don't think we can satisfy just with incremental improvement," he said.

One concept Tenorio floated is architectural rather than incremental: a full decoupling of uplink and downlink, with pervasive deployment of uplink receivers close to the customer, "listening" and relieving the burden on the device. He immediately hedged his own proposal — "It's just an example. Maybe it's not needed. Maybe it's not practical. We'll see about it" — a caveat worth holding onto before treating decoupled 6G as settled doctrine.

What are the other US operators doing?

Decoupling already has traction elsewhere. According to a recent Ericsson blog, the idea is to use different frequency bands in each direction — midband spectrum downstream, perhaps lowband upstream. Rather than splitting a single band between the two directions, decoupling gives each its own "highway," Akhil Gokul, VP and head of technology at Ericsson, told Light Reading at Network X Americas earlier this year.

T-Mobile is examining decoupling as a 6G feature. Egil Grönstad, senior director of technology development and strategy, said at Network X in Dallas that the approach offers "more flexibility on how you're pairing spectrum," letting operators select the best band for each direction rather than working within a fixed ratio. He also raised the possibility of using unlicensed spectrum for added uplink capacity, particularly for fixed applications such as fixed wireless or camera monitoring.

The near-term upgrades are already visible:

• T-Mobile's 5G-Advanced network is rolling out uplink carrier aggregation, uplink MIMO and transmit switching.
• Verizon's 5G standalone network showed marked uplink improvement in a Signals Research Group test after deploying Ericsson's 32T32R massive MIMO radios.
• AT&T cited upstream capacity as part of the rationale for acquiring EchoStar's 600MHz lowband spectrum, and plans to deploy Ericsson dual-band 600MHz radios with "uplink-enhancing 8RX" receiver technology — described as a first for lowbands.

Use cases first this time

Tenorio framed the uplink question as part of a larger correction: with 6G, unlike 5G, the use cases should come before the technical specifications.

"We have the opportunity to design 6G with use cases in mind…as opposed to just specify it from speed and latency and performance, and then hoping that the use cases will make use of it," he said.

To that end, Verizon expanded its 6G Innovation Forum last month with nine new members and highlighted trials in integrated sensing and communication (ISAC) and AI wearables, aimed at pinning down requirements before standards harden.

The timeline is tight. "We have precious two years in front of us," Tenorio said, to answer the open questions and shape a standard that arrives with monetizable use cases rather than after them. Whether the industry finds his missing breakthrough — or settles for incremental uplink gains — will determine how well 6G serves the AI devices Verizon expects customers to wear.